MongoDB OLAP

[bilal berjawi]

Hi,

I would like to compare the performance of an OLAP on a relational
database(Access db) and an OLAP on a NOSQL database. As the first
step, I compared existing NOSQL solutions to choose one of them, I
found that Mongodb is one of the best solutions.
So what I want to know now:

it is possible to use OLAP on top of MongoDB? there's any connectors
that may helps?
if no, what is the alternative to analyse and explore data on MongoDB
like OLAP on relational db?

[Jenna]

Can you give a rough data model, such as the type of data that you'd
like to input and output?

MongoDB isn't designed to work with OLAP cube-type applications, but I
would look into JasperSoft's MongoDB connector:
http://www.jaspersoft.com/jaspersoft-for-mongodb

[NucleonSoftware]

Hi,
Nucleon BI Studio also supports MongoDB.  It provides Reports (Cross-Tab Reports), Charts, Dashboards, Scripts and more features.
For more information:
http://www.nucleonsoftware.com

[bilal berjawi]

hello, 10x for you reply, actually there's no a specific data type, I
want to test the performance of OLAP with a nosql database whatever
the data type is.
I'll check Jaspertsoft. Thank you

[mdahlman]

I can comment on the Jaspersoft story for MongoDB analysis and reporting.

JasperReports Server (both the AGPL community version and the commercial edition) includes an OLAP engine. More precisely it's the ROLAP (Relational OLAP) engine called Mondrian. It's a great engine for OLAP analysis, but it doesn't work with MongoDB. I think it's unlikely to ever work with MongoDB. It relies pretty fundamentally on SQL and joins between tables, and neither of these is likely to be supported in MongoDB soon.

JasperReports Server (both the AGPL community version and the commercial edition) includes a connector for MongoDB. It's great for parameterized reports. They could be run on demand or put into a dashboard or scheduled and emailed out. When run on demand, the reports are interactive. So the end user can sorting, filtering, formatting, rearrange columns, etc.

JasperReports Server (only the commercial edition) also has a drag and drop GUI for analysis and report building. This means that you can build a number of parameterized queries against a MongoDB data source and expose them in this GUI to create charts, pivot tables, etc. This definitely qualifies as "analysis", but I would not call it "OLAP analysis" because it does not use the embedded OLAP engine. It queries MongoDB directly.

Jenna's link to jaspersoft.com will point you to lots more details. There's more on jasperforge.org as well. But the descriptions above perhaps provide enough information so that you can have a feel for whether it would meet your analysis needs.

Matthew Dahlman

Jaspersoft

[bilal berjawi]

Thank you

[bilal berjawi]

Dear Matthew, I checked the JasperSoft and I found it interested but
I've some question, from what I understand that the GUI use the
mongodb connector to create reports,charts,etc but the olap engine
(mondrian) cannot use this connector that's why you would not call it
OLAP analysis right?
So as a conculision, your mongodb connector allows me to create
reports(cross-table),charts... but I can't drill down and roll up and
do things like OLAP.
Did I miss something?

On May 7, 6:22 pm, mdahlman <matt.dahl...@gmail.com> wrote:

[mdahlman]

Yes, I think your conclusion is accurate.

Here's a little extra precision. Remember that you can create a drill-down path from one report to another. So in this respect you do have drill down (and likewise roll-up in the reverse direction). It's not identical to defining an OLAP hierarchy and then letting the OLAP engine handle all drill down and roll up functionality.

Maybe a good example to have in mind is a report in a dashboard showing summary information (the source might be a query that uses the Aggregation Framework to summarize detailed info, or the source might be data that's pre-aggregated in MongoDB). From the dashboard report you can click to drill down to a different detail report which displays all sorts of information at a detailed level (depending on your use case these details might still be aggregates... but at a much finer level than in a dashboard).

Matthew Dahlman 
Jaspersoft